Apple M4 Max (32-core GPU) × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on Apple M4 Max (32-core GPU)
171best gen tok/s
3,347best prompt tok/s
4,352max context tested
9results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_K - Medium | 171 | 3,301 | LocalScore ↗ | 4,352 | 368 ms | llamafile 0.9.2 |
| Q4_K - Medium | 169 | 3,211 | LocalScore ↗ | 512+128 | 371 ms | llamafile 0.9.2 |
| Q4_K - Medium | 167 | 3,325 | LocalScore ↗ | 512+128 | 360 ms | llamafile 0.9.2 |
| Q4_K - Medium | 165 | 3,316 | LocalScore ↗ | 512+128 | 361 ms | llamafile 0.9.2 |
| Q4_K - Medium | 165 | 3,347 | LocalScore ↗ | 4,352 | 364 ms | llamafile 0.9.2 |
| Q4_K - Medium | 165 | 3,234 | LocalScore ↗ | 4,352 | 367 ms | llamafile 0.9.2 |
| Q4_K - Medium | 165 | 3,112 | LocalScore ↗ | 4,352 | 373 ms | llamafile 0.9.2 |
| Q4_K - Medium | 164 | 3,224 | LocalScore ↗ | 512+128 | 375 ms | llamafile 0.9.2 |
| Q4_K - Medium | 162 | 3,197 | LocalScore ↗ | 4,352 | 375 ms | llamafile 0.9.2 |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
"Context tested" is the largest prompt+generation length actually benchmarked (tokens), not a memory-fit estimate.
What else can Apple M4 Max (32-core GPU) run? →
Includes results from LocalScore, a Mozilla Builders project by cjpais, mirrored in the cn0303/local-llm-speed-benchmarks dataset. LocalScore runs use llamafile, which is built on llama.cpp.